Solvency II changes in technical provisions,

slide1. Solvency IIchanges in technical provisions, including contract boundaries Luca Bianchi
May 4, 2023 1<br>
slide2. Introduction EIOPA delivered on 21st April 2022 two guidelines on technical provisions, where the 2nd is focused on contract boundaries, that is the treatment of future premiums and some future options for the point in time technical provisions (pillar 1, reporting).
Why guidelines?
That solution was deemed appropriate in phase of beginning study of “solvency II 2020 review” where some issues were driven to be dealt within guidelines rather than via amendments of regulation.
For this reason, the list of topics in scope of those guidelines is not exhaustive: other can be treated through changes of regulation. 2<br>
slide3. Introduction Final report were delivered on 6th July 2022
Followed by:
- Consolidated version with explanatory text on 10th March 2023
- Revised guidelines on the end of January 2023
IVASS issued on 20th December 2022 document 10/2022 which lists the proposals of amendments of the current local regulation in force IVASS 18, aimed to endorse the changes of the 2 EIOPA guidelines 3<br>
slide4. Divergent practices for technical provisions The need to issue amendments to the existing guidelines has been the observation of divergent practices on the way to calculate technical provisions across Europe during the past years
Therefore, guidelines have been amended with regard to:
The role of future management actions
Modelling future expenses
Valuation of options and guarantees through Economic Scenario Generators
Modelling dynamic policyholder behaviour
Identification of EPIFP

The new guidelines apply in the same manner to both individual and group balance sheets.
Guidelines shall be explicitly accepted by national competent authorities (NCA). If not, EIOPA would inform the EU Parliament, Council and Commission.
Guidelines are subject to potential future changes by EIOPA. 4<br>
slide5. Stakeholders feedback and application date The drafted proposals delivered in April ‘22 were put under discussion to stakeholders which were identified split down in 3 classes:
EU trade, insurance or actuarial associations
National insurance or actuarial associations
Other parties such as consultants, lawyers or insurance and reinsurance undertakings

EIOPA was aware of the importance of first application of IFRS17, considering the scheduled entry in force that is 1st January 2023: it gives to the Local National Competent Authorities (NCAs) the power the delay the 1st application of guidelines 5<br>
slide6. Part 1: technical provisions New: 25, 28A, 37A, 37B, 37C, 40A, 0, 24A, 24B, 24C, 24D, 24E, 40B, 53A, 57A, 77A
Amended: 30, 33, 77 6<br>
slide7. Guideline 25. Modelling biometric risk factors Background
Biometric assumptions in Life Insurance are mortality, longevity, disability.
Disability may be drill down in (1) accident events, (2) LTC events (according to various levels of importance) and (3) Dread Disease.
Pandemic events are biometric events as well, though not usually allowed for setting assumptions for technical provisions.
Guideline content
There are 3 things to pay attention in modelling biometric assumptions:
They have a PDF distribution; however, they are usually modelled through their mean rather than through the whole set of probable outcomes within a stochastic approach. Guideline envisages to consider the possibility to model their stochastic feature
Whenever insurance cover has long term characteristics, the trend of the [mean of] distribution might become important so that the duration of liabilities is key for deciding whether the introduction of changes of parameters in the long term could be appropriate
Biometric risk factors may be correlated each other or correlated with other factors, such as lapses and financial & economic factors. The insurer should assess the correlations. As a consequence, if an economic factor is modelled with different scenarios, the correlated biometric factor should be as well. 7<br>
slide8. Guideline 28A. Investment management expenses Background
Technical provisions allow for expenses which include “investment management expenses” which in turn can be split down in (1) internal, e.g. part of overheads which are classified as “administrative”, (2) external for trading activities and (3) external for managing the assets, for example in looking for the best earning for the investments.
The best estimate expenses should be proportional to the assets under management
Guideline content
Prior to set down the last version of guideline, it was deemed that the new requirement would have compelled to consider future investment expenses proportional to the Solvency II best estimate and SCR. However, the last guideline remains unclear, so that various interpretations may be all good and different practical application would work well.
Anyway, for all the unit linked funds and, for non – linked, those having been ring fenced of where matching adjustment is applicable, the underlying identified assets shall be considered as basis for setting investment expense assumptions. The guideline does not specify whether they should be proportional to either the market value of book value (or to any other measure) because this is left subject to free (expert judgment) entity specific interpretation.
For any other liability, for example liabilities of participating contracts, there’s no particular rule. Anyway, I believe the underlying items as defined under IFRS17 may be appropriate.
At last, reimbursements that the insurer eventually receives by asset managers’ shall be modelled as future inflows rather than as deduction of future outflows that eventually the insurer gives back to policyholders and / or other parties. This assessment is aligned with IAS/IFRS usual requirements, for example IFRS9 applied to financial liabilities. 8<br>
slide9. Guideline 30 (amended). Apportionment of expenses Background
Once identified the volumes which the best estimate investment expenses shall be proportional to (guideline 28), it’s important the identification of parameters and evolution over time of the underlying volumes.
The same work should be done for the whole set of overhead and claim expenses which are in scope of technical provisions.
Guideline content. Parameters
It’s key to project in appropriate way the future expenses to get a realistic and objective forecast.
In order to “allocate overhead expenses over time” the insurer should identify “appropriate expense drivers” and the expense ratios.
In doing so, there should be consistency with the strategic plan and with ORSA (with reference to the next 3-5 years from the valuation date)
In practice, carriers may be the number of contracts in force or the local GAAP net premium reserves for the overhead expenses whereas the claim expenses could be proportional to the liability settled (i.e. annulled as consequence of payment) and / or the number of claims under management. 9<br>
slide10. Guideline 33 (amended). Changes in expenses Background
Once identified the volumes which the best estimate investment expenses shall be proportional to (guideline 28), it’s important the identification of parameters and evolution over time of the underlying volumes.
The same work should be done for the whole set of overhead and claim expenses which are in scope of technical provisions.
Guideline content. Evolution
It’s key to identify the future expenses arising from commitments made prior to the valuation date.
The assessment refers to the need to pay attention to both future expenses arising from future premiums and future options from those stemming from premiums which have already been accounted for in the previous financial statements.
Their evolution (i.e. the value assigned to the parameters in relation to future years, not the volumes) depends on the nature of the expenses involved.
Future expenses shall allow for inflation. Inflation rates shall be consistent (i.e. linked to) with other economic factors
In practice, the inflation rate is linked to the interest rates being modelled with the Economic Scenario Generators. 10<br>
slide11. Guidelines 37A and 37B. Dynamic Policyholder Behavior (1/2) Background
Dynamic Policyholder Behaviour, also known with the acronym PHB, is a characteristic of best estimate liabilities, both under Solvency II and IFRS17, which has its indirect impact in the Solvency II Risk Margin (also in the IFRS17 Risk Adjustment ) as it’s gauged here through shock impacts (through changes corresponding to different quantiles in IFRS17).
It affects as well as the reinsurance recoverable because its risk is charged in part (i.e. mitigated) to reinsurance counterparties independent on the reinsures is charged with a surrender value.
PHB is aimed to measure the changes of basis lapse rates corresponding to different levels of economic factors, other than their historical average, which are deemed to be correlated to.
The basis lapse rates depends usually on the time distance from the origin of contract. For example, if the elderly of contract at the evaluation date is k and the projected year is t, the projected basis lapse rate is a function of time “k+t”.
The historical evidence by means the basic lapses are built up could be associated to levels of financial factors significant different from those as of the valuation date and forecast for future years,.
Thus, even though the best estimate liability is deterministic (i.e. not stochastic), there may be a PHB: the corresponding best estimate is “dynamic and deterministic”
As the financial & economic factors through economic scenario generators and the lapses are linked to their possible outcomes, the corresponding best estimate is “dynamic and stochastic “ . Often, the word “dynamic is hidden” to identify stochastic runs.
The actuarial engine may easily have a switch on / off tool for PHB. The change of best estimate by switching off PHB is the way as to measure the effect of PHB.
To be continued  11<br>
slide12. Guidelines 37A and 37B. Dynamic Policyholder Behavior (2/2) Background
Dynamic Policyholder Behaviour makes lapses raised or decreased depending the relationship between the economic parameter and the lapse rate.
If contracts in force are profitable, the increase of lapse rates in assumptions provokes an increase of best estimate (and risk margin); in contrast, in case of onerous contracts, the increase of lapse rates provokes a decrease of best estimate.
A decrease of lapse rates causes a decrease of best estimate (increase) in case of profitable (onerous) business in force.
Therefore, modelling decreases of lapse rates is important in the same manner as modelling increases of lapses.
PHB is modelled through a link between a variable which expresses the comparison (difference or ratio) of an external economic or financial factor (for example the risk neutral spot rate of an investment with a duration similar to the liabilities) and the interest allocated to policyholders (for example the financial return of the assets of the underlying segregated fund) and the change of basis lapse rate.
The aforementioned variable could work only if its value exceeds a trigger value (meant in absolute value because it works in two directions, the increase and the decrease)
PHB may be multiplicative of additive to the basis lapse rates, depending on the choices of actuarial expert teams of the undertaking.
PHB may be shaped as either linearly or exponentially, depending on the characteristic of the actuarial model. 12<br>
slide13. Guidelines 37A. Dynamic Policyholder Behavior Guideline content
Dynamic Policyholder Behaviour must be allowed for in the calculation of technical provisions.
To identify its measure, the entity should use statistical tools on the base of empirical evidence
As alternative, expert judgment is permitted, however, with the help of clear documentation.
Since PHB is more evident under extreme [economic] scenarios, the lack of historical empirical evidence between those extreme [economic] events and lapses “should not be considered alone to be a reason to avoid dynamic policyholder bahaviour modelling”
There’s interaction between PHB and future management actions
The last assessment is of a key importance in life insurance: it could easily make the best estimates raised in scenarios whose management actions make the future earning of declined.
The more lapse rates in extreme scenarios when projecting profitable business in force, the higher the best estimate. 13<br>
slide14. Guidelines 37B. Bidirectional assumptions Guideline content
Dynamic Policyholder Behaviour must be modelled in two directions: the increase and the decrease.
There are no further indications on how the two different directions could play, hence modelling is left free under discretion of actual expert teams of the Undertaking.
Increase of decrease shall depend on the variable described in my previous pages that is referred to as “trigger event”, not to be confused with the word “trigger” meant as the minimum level of the variable beyond that the PHB is working. 14<br>
slide15. Guidelines 37C. Option to pay additional or different premiums Background
Contracts often offer the to the policyholders the option to pay additional premiums or to vary the recurrent premium, in its amount or in the frequency of instalments.
Single premium contracts offer as well as the option to pay additional premiums (top – ups) even though not necessarily all the products supply this kind of option.
Should contract boundaries on future premiums be not applied, the actuarial model should project foreseeable premiums and allow for these and the relevant obligations & expenses in to the point in time technical provisions.
It does not matter whether future premiums are onerous or profitable: their inclusion is not dictated by need to be prudent.
Remind: best estimate are realistic!
Guideline content
As for top – ups, their contribution shall be part of model for technical provisions.
As for recurrent premiums, the actuarial model shall foresee the possible trend of their variation over time. 15<br>
slide16. Guidelines 40A. Comprehensive management plan Background
Management actions applicable in future must be taken into account to project future cash flows. The requirement ensures the realism of technical provisions.
Changes of management actions are permitted in the actuarial model in response of shocks such those considered in the SCR.
Both should be documented and approved by the Board: since then, their mathematical application become area of actuaries and CFO.
Generally, the ALM used to project future earnings of the underlying assets, in conjunction with dynamic policyholder behaviour, makes wide use of management action for the valuation of best estimates as well as the cash flows under shock (stress) make wide usage of changes of management actions.
Even in phase of valuation of non – stressed best estimate face extreme scenarios in 2% - 10% of cases of the Economic Scenario Generators: changes of management actions are applicable in these scenarios.
Guideline content
Future management actions must be stored either in a single document or in a set of documents accompanied by an inventory.
Those documents shall list the management actions in order to let users get a complete view. 16<br>
slide17. Guidelines 0. Proportionality Background
Technical provisions should be modelled through actuarial models and run at least every quarter whose outcomes fill in the quarterly reporting.
There are plenty of refinements making different ranking about the quality of formulas and accuracy of calculations
Proportionality is a general concept about accuracy and quality. Where the business is not material inside the Undertaking, the actual model may give up some accuracy.
Proportionality is closely related to the relative limits of acceptance of technical provisions in terms of accuracy.
Guideline content
Proportionality is tied to the need technical provisions should not deviate from the price of liabilities set in case of transfer of portfolio.
It’s worth noting that reference is the part of portfolio (business) which the undertaking intends to apply proportionality to, not the whole portfolio in force. 17<br>
slide18. Guidelines 24A. Use of expert judgment (1/3) Most of usage of expert judgement in setting technical provisions is shared with their use in internal models
The undertaking have more data available to run best estimate liabilities than to run the SCR for internal model purposes, hence the requirements about expert judgement (briefly EJ hereafter) is slightly different.
EJ, where applicable, needs to implement a framework for
Documentation
Governance
Validation
Both governance and validation involve:
Sharing and approval process in internal committees and, depending on materiality
Senior executive committees or even the Board 18<br>
slide19. Guidelines 24A. Use of expert judgment (2/3) Sensitivities
Performing sensitivities to some material assumptions which drive the value of best estimate is the main way for bearing the use of EJ
The outcome of any sensitivity should be a percentage of impact, i.e. relative variation & change of either technical provisions or expected value of future profits (VIF).
Furthermore, in order to understand materiality, is of key importance to understand whether the outcome of sensitivity is able to change or influence (should it become real) the program of reinsurance or the underwriting / selling of insurance products or even the decisions of Board.
The EJ validation shall be embedded in the process of validation of assumptions: it’s does not make sense that different internal committees discuss or approve modelled assumptions and unmodelled or uncertain assumptions which require EJ
Binary events and extreme scenario
Guideline 24A speaks about the need toc heck the impacts of binary events and of extreme scenarios in the process of setting assumptions
Examples are:
Environmental issues such as global warming
Legislative or political changes
to be continued  19<br>
slide20. Guidelines 24A. Use of expert judgment (3/3) Factors which are able to modify the claim amounts (raise) or the volumes of new business (decline)
These factors mirror the actual situation occurred since Q1 2022: the increase of interest rates and inflation rate which is driven by economic and financial factors whose root is in part the war in Ukraine, that is a political change
It’s worth noting that, although future new business is not allowed for having a direct impact on the point in time technical provisions, they have indirect impacts both through the expense assumptions and via the ALM, i.e. the future discretionary benefits, that is the PHB, TVOG and the FDB.
EJ is also present in deciding the “weight to be placed on events that are not in the [underlying] data [for setting assumptions].
Conclusion
Guideline 24A speaks about the need of
Take into account materiality in EJ
Usage of both quantitative and qualitative key indicators to understand materiality
Usage of binary events and extreme scenarios to understand materiality
Usage of events that are not present in the entity specific historical data 20<br>
slide21. Guidelines 24B. Governance of assumptions setting The setting of assumptions and EJ follow a validated and documented process
Derivation and usage: they must be consistent
over time and
across the Undertaking
Assumptions must be fit for their intended use
Escalation:
Depending on their materiality, the approval of assumptions shall be done at different levels of seniority, even up to the supervisory body or to the Board

“Derivation” includes the data underlying statistical analyses and the way to perform the statistical analyses 21<br>
slide22. Guidelines 24C. Communication and uncertainty in assumptions setting As regards the process of setting assumptions
The way of choosing assumptions and particularly the relevant EJ must be involve different and independent roles inside the Undertaking.
The Undertaking shall ensure there’s no misunderstanding of miscommunication between them
For example, when lapse assumption are dealt with reference to surrenders, everybody should understand whether discussion is focussed on portfolio transfers (individual pension funds such as PIP) or reimbursement of term insurance linked to mortgage and loans which depend of ceased mortality risk (the mortgage or the loan has extinguished) rather than to a surrender of participating / unit linked contracts whose underpinning motivation of policyholders is financial.
Users and providers of assumptions are separated and supply different roles.
The Undertaking shall ensure there’s a “formal and documented process between providers and users” of EJ and assumptions
It’s worth noting that actuarial function does coincide with neither users or providers, rather they are different parties by means 1st line operates and actuarial function is another (third) party with control function.
The level of uncertainty of assumptions has to be transparent and the consequents “variation” in final results as well.
The “variation” is meant to be the impact in the last or in the next reporting period. This is not a sensitivity. I believe the word includes, not only changes of technical provisions, also changes of Own Funds and changes in the SCR due to the feature that underwriting risks are valuated via shocks of best estimates. Also, future profits stemming from future new business and used (under a shock environment) to bear the Loss Absorbency Capacity of deferred taxes (LACDT) would be affected. As consequence, an info about the impact in solvency ratio may be useful 22<br>
slide23. Guidelines 24D. Documentation of assumption setting As regards the process of setting assumptions
Similarly to guideline 24B, guideline 24D speaks about the need to provide documents on the assumption setting and on EJ, adding that it should ensure transparency.
As regards the document describing the assumptions
Document shall unveil the following:
The Materiality
The experts engaged
Intended use
Period of validity

Process of decision making has to be transparent e supply details aimed to understand:
the rationale of opinion,
the information used to issue the opinion,
decision making criteria to select assumptions,
the alternative setting rejected (deemed to not be appropriate)

As for material assumptions
They shall be documented in a written way in order to be clear and understandable 23<br>
slide24. Guidelines 24E. Validation of assumption setting (1/2) As regards the process of choosing assumptions
Process of choosing assumptions and EJ has to be validated.
As regards the process and the tools
All these must be documented
Track changes for material assumptions
Material assumptions should be tracked through experience variance exercises.
Material assumptions should be tracked in case of new information could modify their setting
In both the cases, analyses explaining the changes are necessary
Tools: stress testing and sensitivities
Stress testing and sensitivities are strongly advised as tools of validation

to be continued  24<br>
slide25. Guidelines 24E. Validation of assumption setting (2/2) Independent review
Internal or external reviews of experts who guarantee independency should be appropriate.
It’s worth noting that e actuarial function is deemed to be independent and fit for purpose, hence fulfilling the requirement

Subsequent invalidation of assumptions
Assumptions are based on historical experience, both entity specific and internal.
New information about factors affecting assumptions could invalidate their appropriateness.
Not only the Undertaking has to check the circumstances able to compromise the suitability of assumptions, it should “detect” the circumstances in advance whose occurrence would make assumptions false. 25<br>
slide26. Guidelines 40B. Consideration of new business in setting future management actions (1/3) Background
This is the most controversial new guideline.
Future management actions are not assumptions, they are something relevant to the management of assets underlying the liabilities.
Not only the asset mix chosen to optimise the yields and the risks in accordance with local regulation IVASS 24, also the way and time as to realise gains and losses before maturity of bonds are part of strategic asset strategy
The investment strategy to be followed and the reaction to extreme external financial scenarios are key to predict asset management actions.
It’s clear that those strategies depend on the expected volumes which in turn depend on the forecast new businesses and on future premiums foreseen to originate from the business in force and which are not yet contributing to technical provisions due to contract boundaries being applicable.
However, the joint projection of future obligations stemming from past premiums and future premiums outside contract boundaries and – at the same time – future new business and future premiums which meet contract boundaries would generate strange effects on the expected yields of the underlying assets, particularly in respect to participating contracts.
In periods of increasing interest rates, the ALM tool in usage for the stochastic valuations would predict higher earnings due to the positive contribution of future investments of future businesses;
In contrast, in period of decreasing interest rates, the ALM would predict lower earnings due to the negative contribution of future investments of future businesses
to be continued  26<br>
slide27. Guidelines 40B. Consideration of new business in setting future management actions (2/3) Background
If so (future new business explicitly modelled in the ALM), the consequences would be:
An impact in the liability for future discretionary benefits and in the loss absorbency capacity of technical provisions
An impact on TVOG, the stochastic cost of financial guarantees and of future discretionary benefits
An impact driven by the PHB of contracts in force as reaction of changes of expected yields driven by future new businesses
A partial double counting with future profits arising from future new businesses which, under a stress scenario, support the deferred tax assets posted in the balance sheet or, alternatively, support the loss absorbency capacity of deferred tax assets (LACDT)
The double counting relates to the contribution of future investment income to those profits which, in turn, should be unbundled from the contribution of underwriting profits (not practicable due to the interaction of PHB).
The same considerations are valid when we look at future premiums of business in force which meet contract boundaries
Therefore, we would work on the ground to not allow for the contribution of future cash flows stemming from future new business (and from future premiums now subject to boundaries) to the expected yields of the segregated fund.
Instead, we would work on the ground that future new businesses (and premiums & cash flows arising from contracts in force whose premiums are tied by contract boundaries at the reporting time) have a contribution through the strategies to manage the assets in force which underly the technical provisions as of the valuation date (the time of reporting)
to be continued  27<br>
slide28. Guidelines 40B. Consideration of new business in setting future management actions (3/3) Guideline
It’s reported in full text below:
“Insurance and reinsurance undertakings should consider the effect of new business in setting future management actions and duly consider the consequences on other related assumptions. In particular, the fact that the set of cash-flows to be projected through the application of art. 18 of the Delegated Regulation on contract boundaries il limited should not lead insurance and reinsurance undertakings to consider that assumptions only rely on this projected set of cash-flows without of any influence of new business. This is particularly the case for the allocation of risky assets, management of the duration gap or application of profit sharing mechanisms”

The last sentence “profit sharing mechanisms” may refer to general rules for allocating interests to policyholders so that to avoid explicit references to the contribution of future investment income stemming from future businesses to the yields of assets for the valuation of point in time best estimates.
The wording will be used also in the next new guideline 53A 28<br>
slide29. Guidelines 53A. Use of stochastic valuation (1/3) Background
“Stochastic modelling is appropriate whenever future cash flows in the scope of technical provisions depend on future events and developments, in particular those with material options and guarantees”
It’s worth noting that these conditions incur not necessarily in respect to financial and economic assumptions.
Therefore, underwriting factors are embraced in the general requirement of stochastic modelling
The usage of stochastic run make sense in presence of asymmetry between results by varying one identified assumption.
The most common case is represented by financial guarantee which is often supplied within participating contracts and sometimes within unit linked contracts.
If so, looking at several scenarios of returns of the underlying assets forecast in run off (i.e. for at least 40 years beyond the valuation date), the average roughly the stream corresponding the mean being used in the deterministic run (i.e. non – stochastic).
Thus, why stochastic run in respect to different scenario di assets returns should modify the best estimates and why, in particular, the best estimate should raise?
The answer is twofold:
there are scenarios where financial guarantees become in the money, that means the revenue proportional to the expected management fee on the underlying assets cannot be earned (net of part which the sale network or other counterparties would waive to) anymore or are earned to a limited extent
To be Continued  29<br>
slide30. Guidelines 53A. Use of stochastic valuation (2/3) Background
The answer is twofold:
Future discretionary benefits become extreme in some scenarios (very high) which do not match the market value of assets posted in the balance sheet at the reporting date. Therefore, the difference between the stochastic and the deterministic FDB is generally >0 whilst, in contrast, the market value of underlying assets accounted for in the balance sheet is just one single amount, not deterministic.

Both the stochastic cost of financial guarantees and the stochastic cost of future discretionary benefits depend on how PHB works within the stochastic ALM.
Generally, the PHB has an adverse outcome, i.e. let the best estimate increase, despite its bidirectionality because, looking at profitable business in force, the adverse impact of scenarios where lapses go up is only partially offset by the favour impact of scenarios where lapses go down.
It’s worth noting that scenarios of assets returns are generally not achieved through direct simulation of yields, rather they are calculated with analytical formula by simulation of the economic parameters which are in close relationship with the assets returns.
They are, for examples, the interest rates, the spread, the exchange rates, the equity index, dividend yield and, of course, the inflation rate. 30<br>
slide31. Guidelines 53A. Use of stochastic valuation (3/3) Guideline
To assess whether stochastic modelling is needed, the Undertaking shall understand if the model captures the value of options and guarantees.
Note that “options” in life insurance are at least
The conversion of lump sum in annuity or vice-versa the conversion of an annuity in lump sum at the end of vesting or accumulation period
Deferral of maturity date for products other than whole life
Voluntary switches, particularly from participating to unit linked or vice-versa in hybrid products
There are other feasible options, starting from the lapse with surrender value and the payment of additional premiums: all these may be under observation in a stochastic model
In reasoning about the need of a stochastic model, the undertaking has to pay attention to:
Any kind of profit-sharing mechanism where the future benefits depend on the return of assets
Financial guarantees (e.g. technical rates, even without profit sharing mechanism), in particular, but not only, where combined with options (e.g. surrender options) whose dynamic modelling would increase the present value of cash flows in some scenarios.
The last sentence identifies the concept of prudence: it’s really important the usage of stochastic modelling if the interaction of PHB and financial guarantees make the best estimate higher, hence in theory not necessary in the opposite case. 31<br>
slide32. Guidelines 57A. Market risk factors needed to deliver appropriate results Guideline
The model of risk factors [through ESG] shall reflect the following:
The volatility of assets, including spreads & default
Negative interest rates 32<br>
slide33. Guidelines 77 (amended). Assumptions used to calculate EPIFP (1/3) Background
The best estimate may be split down in (1) those relevant to past and current premiums and (2) those stemming from future premiums .
For most of contracts at recurrent premiums and contracts at single premiums with the policyholders’ option to pay additional premiums, the division of obligations attributed to past and current premiums from those attributed to future premiums is easy.
The fact that expense assumptions on future premiums and on past & current premiums are influenced each other is not a concern for the task to split down expenses in two pieces because we haven’t to simulate that policyholders interrupt the payment of future premiums to gauge the best estimate of past and current premiums
When the sum assured is a combination of past, current and future premiums which is the characteristic of contracts sold until the ‘90s of past century, we need to simulate the effects of the interruption of premium payment at the reporting date. That interruption is referred to as “paid up” [event].
The formula of paid up permits to determine objectively (for the sake of clarity o policyholders) the reduction of sum assured in case of live at maturity and / or in case of death. After that, the calculation of the mathematical reserve (net premium reserve, best estimate) is consequential.
It’s rare that the event of paid up originates directly a change of mathematical reserves and the sum assured is then refreshed.
The sum assured is often subject to penalization. It does matter what kind of paid up penalty is applicable, the guideline says “… the calculation should not include penalties, reductions or any other type of adjustment to the theoretical valuation of technical provisions without a risk margin calculated as though the policy continued to be in force” 33<br>
slide34. Guidelines 77 (amended). Assumptions used to calculate EPIFP (2/3) Background
Apart penalties, there are possible changes in other parameters should paid up become reality
For example, the Undertaking could argue that policyholders who will waive to pay future premiums have a different mortality compared to policyholders who are continuing to pay future premiums.
The lapse rates may be different as well: for example, who won’t pay any longer future premiums could be subject to higher surrender rates than policyholders who are continuing to pay premiums
Last example: the probability of the option to convert the lump sum in life contingent annuity might change after paid up.
For all these cases, the guideline is clear:
All the other assumptions (e.g. mortality, lapses or expenses) should remain unchanged [when simulating the interruption of payment of premiums from the reporting date onwards]. This means that the insurance and reinsurance undertakings should apply the same projection horizon, future management actions and policyholder option exercise rates used in best estimate calculation without risk adjusting them to consider that future premiums will not be received. Even if all assumptions on expenses should remain constant, the level of some expenses (e.g. acquisition expenses of investment management expenses) could be indirectly affected. 34<br>
slide35. Guidelines 77 (amended). Assumptions used to calculate EPIFP (3/3) Future management actions unchanged entails that future bonus rates must kept constant, i.e. the ALM projections of earnings of the segregated funds underlying participating business shall remain the same.
Of course, the liability for future discretionary benefits and the TVOG would change since proportional to different volumes of liabilities under management.
The guideline says that the only changes admitted relate to the acquisition expenses and the investment management expenses:
Acquisition expenses are meant to be both the recurrent commissions and the part of management fees earned on the volumes which are withdrawn to the sale network and to other counterparties
The commissions fees paid to investment managers should be modified as well
The rationale behind the possibility to modify those cashflows is either the absence of future premiums (this is the aim of simulation), which entails that no explicit commissions on premiums will be due, or the reduction of volumes of investments stemming from the assumptions no investments arise from future premiums, which entails that the corresponding management fees will not be paid to any kind of counterparty.
It’s worth noting the absence of any requirement about the allowance of gross management fees. This is due to the fact that gross management fees are not explicit cash inflows of the best estimates; it’s a consequence of the general rule about the prohibition to recognize the expected future investment income in the valuation of best estimates. 35<br>
slide36. Guidelines 77A. Alternative approach to calculate EPIFP As simplification, the EPIFP may be indirectly valuated as the share of present value of future profits (VIF) stemming from future premiums.
The guideline in draft foresaw a sign off by the actuarial function, anyway the last guideline does not mention any sign off as necessary, 36<br>
slide37. Part 2: contract boundaries New: 0, 6A, 6B, 6C
Amended: 5 37<br>
slide38. Guidelines 0. Contract boundaries (1/2) Background
The reason of the issue of guideline 0 is the (few) Undertakings’ bad practise to annul the contribution of cash outflows falling in future periods which go beyond any contract boundary decided for any future premium!
They had annulled the contribution of those cash flows despite are guaranteed by past or current paid in premiums.
The equation of best estimate identifies cash flows as function of “t” and “k”
“t” is the time beyond 0 (zero) that is conventionally the evaluation date (and the reporting date as well in pillar 1)
“k” is the premium by means the cash flows out had originated (k<=0) or will be originated (k>0).

For any k>0, the best estimate equation allows for both the cash flows out and the future premiums (cash flow in) by means the first are justified
If there’s a contract boundary at time k’, the consequence is that neither premiums nor the relevant obligations from k>=k’ must be projected at the valuation date 0.
However, if this is the case, the obligations stemming from premiums paid in k<=0 and premiums foreseen to be paid at any 0<t<k’ must be allowed for in the best estimate in 0, for any t>=k’.
Unfortunately, some Undertakings have (wrongly) interrupted the consideration of future obligations at time t=k’ so that only obligations in t<k’ have a contribution to the best estimate in 0 38<br>
slide39. Guidelines 0. Contract boundaries (2/2) Guideline
The guideline tries to explain that the right method is that corresponding to the blue one of my previous slide whilst the wrong method is that corresponding to the red one.

Furthermore, the explanatory text says:
If the undertaking can compel the policyholder to pay the (future) premium, contract boundary is not applicable (i.e. is not met)
If the policyholder can pay the (future) premium which the Undertaking cannot refuse to, the contract boundary is not applicable
If either the policyholder or the undertaking could, respectively, asks for receiving back a premium already paid in and refuse to accept the premium already paid in, for - both - during a short period of time, then in practise contract boundary is not applicable. 39<br>
slide40. Guidelines 5 (amended). Contract boundaries (1/2) Guideline
The explanatory text of guideline explains that, for the purpose of setting contract boundaries about options and guarantees, any contract which could be detached in two different independent contracts, need to be unbundled because, if not, “the contract could lead to different contract boundaries compared to independent contracts,….”
“therefore, any contract that is equivalent in terms of risk to two (or more) parts of the contract [which] could be sold independently should be unbundled”
“Conversely, dependencies in term of risk should prevent a contract from being unbundled unless they are not discernible”
Examples where unbundling is necessary
Unit linked whose mortality cover is the sum of fixed amount and a % of face value of units
Saving products with different components where the allocation of premiums is under the policyholder’s discretion
Saving products with an additional mortality cover whose cost is borne by policyholders’ through a predefined rider 40<br>
slide41. Guidelines 5 (amended). Contract boundaries (2/2) Guideline
Examples where unbundling is not necessary
Unit linked whose mortality cover is the maximum of fixed amount and the face value of units
Saving products with different components where the allocation of premiums or reserves is under a dynamic reallocation that is neither controlled by policyholder nor predefined
Saving products with an additional mortality cover whose cost is borne by policyholders’ through a rider which can be re-priced and changed by the Undertaking
Unit linked products with an additional mortality cover whose cost is deducted periodically from the unit fund 41<br>
slide42. Guidelines 6A, 6B, 6C. Identification of a financial guarantee of benefits, insurance risks with a discernible effect on the economic of a contract and changes over time Background
If the contract has no discernible impact from the point of view of Undertaking, contract boundary is met, i.e. it’s applicable.
The underpinning rationale is that events that do not change technical provisions and do not change the SCR or, more widely, risks, may be excluded from the point in time best estimate.
Of course, where premiums are due mandatory, contract boundary is not applicable even though without discernible effects.
The existence of any discernible risk should be identified at the inception (origin) of contract, however changes over time are feasible.
It’s worth noting that contract boundary is met also in the opposite event where the Undertaking is strongly exposed to risks and, nevertheless, can reject or amend prices & conditions to mitigate those risks.

Guideline 6A deals with financial discernible risks, guideline 6B deals with insurance discernible risks and guideline 6C deals with changes of risks after the origin of contracts.
To verify the materiality of any future option or future premium, a quantitative check is envisaged. It consists in excluding the financial guarantee or the insurance cover in the best estimates and gauge the difference; the best estimate to be compared must allow for the the option or future premium under investigation.
Thus, the materiality assessment represents a sensitivity test where future obligations arising from (1) past and current paid in premiums, (2) future premiums deemed to not meet contract boundaries and (3) future premiums deemed to meet contract boundaries (the last being under investigation) contribute all together to the results without distinction between (1), (2) and (3). 42<br>
slide43. Guidelines 6A. Identification of a financial guarantee of benefits with a discernible effect on the economic of a contract Guideline 6A
To determine whether a financial risk has a discernible effect (from the point of view of Undertaking), its impacts shall be seen in respect to all the future cash flows.
Financial guarantees, in particular, drive to existence of discernible effects since provides the policyholder with a discernible advantage.
The check maybe either quantitative or qualitative
Under a qualitative check, the Undertaking should look at the timing, amount and risks arising from the financial guarantee.
Under a quantitative check, the Undertaking should run the best estimate with a without the guarantees. In doing so, they have to pay attention to 3 things:
Run stochastic calculations, at least via closed formula. In case of scenarios, look at the average of outcomes
Compare the best estimate with and without financial guarantees excluding the volatility adjustment and the matching adjustment from both the best estimate runs
Consider the impact of future discretionary benefits if and only if they are in scope of best estimates, i.e. if they are not “pure” discretionary benefits
The last sentence means that FDB is considered to the same extent as financial guarantees 43<br>
slide44. Guidelines 6B. Identification of a coverage for a specified uncertain event that adversely affects the insured person with a discernible effect on the economic of a contract Guideline 6B
To determine whether an insurance risk has a discernible effect (from the point of view of Undertaking), its impacts shall be seen in respect to all the future cash flows.
Insurance cover (including guarantees of minimum in case of biometric event) in particular, drives to existence of discernible effects since provides the policyholder with a discernible advantage should the insured (biometric) event incur to the head insured.
The check maybe either quantitative or qualitative
Under a qualitative check, the Undertaking could look, for examples, to the % of additional cover or to the level of price of biometric cover. More in general, it should look at the timing, amount and risks arising from the insurance cover
Under a quantitative check, the Undertaking should run the best estimate with a without the biometric cover. In doing so, they have to pay attention to 3 things:
Run stochastic calculations when necessary. In case of scenarios, look at the average of outcomes
Compare the best estimate with and without insurance cover including the volatility adjustment and the matching adjustment from both the best estimate runs
Despite expenses are not insurance cover, the quantitative check shall pay attention also the impact on the best estimates for future expenses 44<br>
slide45. Guidelines 6C. Reassessment of a discernible effect of a cover or financial guarantee. Guideline 6C
Once defined whether or not contract boundaries are applicable for each future premium and for each future option, to modify the stance is a rare event according to Guideline 6C.
If the Undertaking would decide in favour of change, it must declare the choice suddenly to the National Supervisor and disclose the rationale in the next reporting RSR if due by the end of year, otherwise with a report ruled by art. 312 (3) of EU Regulation 2015/35 (10 Oct 2014), “including a detailed description of the reassessment and its impact on the solvency position of the Undertaking”.
The change shall be originated from external event such as changes of interest rates as of those existing at time of inception of the contract (i.e. when decisions about contract boundaries were taken).
If due to interest rates, any change less significant than the previsions of interest rates up and down of Standard Formula should not be considered. The changes must be extreme to justify the reassessment of contract boundary.
The materiality of change since contract inception is checked via impacts on technical provisions. In doing so, the changes due to the extreme scenarios of stochastic runs should not be considered; moreover, impacts on stressed scenarios used to calculate the SCR should not be considered.
The reassessment of contract boundary follows the same route than at inception, in particular it’s stated on the basis of the average expected value of future possible cash flows and, in doing so, for the non - stressed best estimate.
Thus, once defined (or reassessed), any contract boundary keeps unchanged (i.e. constant) under extreme scenarios being used for the calculation of best estimate as well as keeps constant under the stressed scenarios used for calculating the SCR sub-modules. 45<br>